The MonitorShrimp project, carried out from 1 January 2021 to 30 June 2023, was coordinated by the Alfred‑Wegener‑Institut of the Helmholtz‑Centre for Polar and Marine Research (AWI). Its goal was to create a prototype system for monitoring the shrimp species *Penaeus vannamei* in recirculating aquaculture systems (RAS) under commercial conditions, with a focus on animal welfare parameters such as stocking density, growth, mortality and stress. The project was funded by AWI and involved three main partners: MonitorFish GmbH (which was dissolved during the project), Erwin Sander Elektroapparatebau GmbH (ES), and Oceanloop Kiel GmbH & Co. KG (OLK), formerly known as Förde Garnelen. Oceanloop’s representative, Dr Bert Wecker, served as the on‑site coordinator, while Dr Matthew James Slater of AWI provided overall project coordination. A subcontractor, NeuroSYS, was engaged to develop the image‑capture application that feeds data to the central server.

From a technical standpoint, the project addressed the lack of automated, non‑intrusive monitoring tools for shrimp welfare. Initial attempts with a camera system developed by MonitorFish proved inadequate under the low‑light, high‑temperature, and humid conditions typical of tropical shrimp farms, and the system’s IP rating was insufficient to withstand salt‑laden aerosols. Consequently, the team tested a range of commercial cameras, including high‑end models used in automated processing lines, but these were rejected due to cost and only marginal image‑quality improvements. The breakthrough came with the adoption of a modern smartphone camera. Smartphones offer a high‑resolution sensor, advanced image‑processing algorithms, and robust protective housings, delivering a cost‑effective solution that significantly improved the accuracy of shrimp counting and length measurement. The system achieved a counting accuracy of at least 95 % and a length‑measurement accuracy of 95 %, meeting the project’s performance targets. Moreover, the image‑analysis pipeline, developed by NeuroSYS, automatically detected color changes in the shrimp’s tail fan—a reliable visual stress indicator—allowing real‑time welfare assessment.

To generate training data for the welfare‑alert algorithm, the team induced stress by temporarily increasing stocking densities in isolated sections of the production tanks. This approach ensured that 10–20 % of the shrimp displayed observable stress symptoms, providing sufficient examples for the machine‑learning model. In parallel, hydrophones installed at the AWI Aquaculture Research Centre recorded acoustic signatures of shrimp behavior, which were later correlated with visual data to refine the welfare detection model. The software stack also included a server‑side analytics engine that processed the images, extracted key metrics, and generated alerts for farm operators, enabling rapid intervention and reducing mortality.

The collaboration structure facilitated a rapid transition from concept to prototype. MonitorFish supplied the initial camera hardware and conceptual design, while ES contributed expertise in electrical appliance construction and integration of the imaging system into the RAS environment. Oceanloop provided the commercial shrimp farm setting, allowing real‑world testing under commercial production conditions. NeuroSYS handled the development of the mobile application and server‑side algorithms, ensuring seamless data flow from field to analytics. AWI coordinated the overall project, provided research infrastructure, and managed the integration of acoustic and visual data streams.

In summary, the MonitorShrimp project delivered a proof‑of‑concept system that automatically monitors shrimp welfare with high accuracy, reduces handling stress, and enhances production efficiency. By combining affordable smartphone imaging, advanced image‑analysis algorithms, and acoustic monitoring, the project offers a scalable solution that can improve the competitiveness of German shrimp aquaculture and support sustainable seafood production.